1. Field of the Invention
The present invention relates to high efficiency power amplifiers and more particularly to a new class of switching power amplifiers that is a hybrid of class E and inverse class F (class Fxe2x88x921) power amplifiers.
2. Description of the Related Art
Power amplifiers are classified in several different categories such as A, AB, B, C, D, E, F, S, etc. based on their fundamental characteristics, which relate to circuit topology and principle of operation. Each class presents relative advantages and disadvantages in their operating characteristics, such as linearity, power efficiency, bandwidth, frequency response, etc., and is chosen according to the application requirements.
More particularly, RF power amplification can be realized using active devices (i.e. transistors, vacuum tubes), that function as linear amplifiers, switching amplifiers or as a combination of both. Since linear amplifiers (e.g. classes A and B) are relatively inefficient at producing radio frequency (RF) output from an applied signal and direct current (DC) supply power, operating an active device as a linear amplifier is not an ideal solution for power amplifier applications requiring high efficiencies. Rather, designing the active device to operate as a switch is preferred because this mode of operation causes the device to be in a saturated or cut-off condition most of the time and therefore dissipates relatively little power by keeping the device out of the much lossier active region. In many applications, such as portable communication devices (e.g. cell phones) and high-power industrial generators (e.g. plasma drivers and broadcast transmitters), where low power consumption and low dissipation are crucial, high efficiency switching amplifiers are an attractive solution due to the performance and cost advantages they allow.
FIG. 1 simplified block diagram of a generic switching power amplifier 6 designed into a conventional RF transmission system 1. The system includes a driver 4, the power amplifier 6, comprising a switch 5 and load network 7, and a load 8. The input signal 2 to be amplified is input to the driver stage 4, which controls the active device 5 in the amplifier. The active device acts substantially as a switch when appropriately driven by the driver and thus is represented as a single-pole, single throw switch. The active device is powered by a dc power supply 3, and has an output connected to the input of the load network 7. The output of the load network 7 is connected to the load 8, such as an antenna. As the switch 5 is cyclically operated at the desired output frequency, or fundamental frequency, fo, the dc energy is converted into ac energy at this switching frequency and its harmonics. The load network 7 may employ one or more filters to control the power dissipation caused by switching action (i.e. the efficiency of the device), reduce the level of the harmonic overtones at the load, and/or provide impedance transformation. The design of the load network determines the behavior of the voltage and currents in the switching amplifier 6, and thus the class of operation by which the amplifier is denoted.
Realizing highly efficient switching operation at high frequencies, however, has been challenging due to finite switching times in the active device and package parasitic impedances. Nonetheless, among the known types of power amplifiers, when an application requires highly power-efficient amplification at high operating frequencies, ostensibly the most appropriate known types are class E and F amplifiers.
Class E Amplifiers
The class E amplifier achieves high efficiency at high frequencies by essentially eliminating the dominant cause of the switching power dissipation that occurs in other types of switching amplifiers, namely the loss associated with capacitive discharge. In virtually every switching-mode power amplifier, a capacitance, CS, shunts the power switch. At a minimum, this capacitance is the inherent parasitic capacitance, Cout, of the circuit components (transistor) and wiring; the circuit designer might intentionally wish to add additional capacitance. In other types of switching amplifiers (other than the Class E amplifier), this shunt capacitance is typically undesirable. The reason is that if the switch is turned on when the voltage across the switch and its shunt capacitance is nonzero, the energy stored in the charged capacitance will be dissipated as heat; the energy is CsV2/2, where Cs is the capacitance shunting the switch and V is the voltage across the switch (and hence across the capacitance) when the switch is turned on. If the switching frequency is fo, the power dissipation is CsV2 fo/2. Note that the power dissipation is directly proportional to the switching frequency. Thus, for a high-frequency power amplifier, this power dissipation can become a severe drawback, often becoming the dominant power loss mechanism. Moreover, while the switch is discharging this capacitor, the switch is subjected to both the capacitor voltage and the discharge current, simultaneously. If the simultaneous voltage and current are large enough, they can cause destructive failure and/or performance degradation of the power transistor.
These difficulties can be avoided by ensuring Zero-Voltage-Switching (ZVS) operation, i.e. demanding that the voltage across the switch be substantially zero when the switch is turned on. This feature of the class-E amplifier allows this class to readily accommodate the switching device output capacitance without seriously degrading performance by using this capacitance in the load network and designing the load network so that the capacitor voltage is zero at just before the device turn-on.
In addition to the problems with turning on the switch, switching off (opening) a power switch inherently subjects it to simultaneous high voltage and high current (hence further power dissipation and device stress). Fortunately, unlike the turn-on loss, this loss mechanism can be made arbitrarily small by choosing a faster device or increasing the device drive level sufficiently so as to reduce the device turn-off time. Although it is possible to design a switching amplifier to achieve ZCS (zero-current-switched) operation, wherein the device current is zero just before the transistor switches off thereby eliminating turn-off loss, it is believed to be impossible to achieve ZVS and ZCS conditions simultaneously. While the turn-off loss can be reduced by other means, the turn-on loss is dependent only on the switching voltage and the capacitance, CS, which cannot be reduced arbitrarily as it is an inherent property of the active device. Therefore, ZVS switching has been found to be the most appropriate for high-efficiency operation using modern high-speed devices. By properly choosing the relative values of the circuit components (including the switch capacitance CS, the load resistance RL, and load inductance LL), class E therefore allows for ZVS switching to reduce switching loss using a very simple circuit.
Thus, with relatively simple circuit topology, class E operation achieves low power dissipation and low device stress by (a) incorporating the switch shunt capacitance as part of a network, allowing its detrimental effects to be accounted for and minimized and (b) using a resonant load network whose transient response after the switch turn-off brings the switch voltage back to zero (or nearly zero) at the time the switch will next be turned on. A schematic of a typical class E amplifier circuit is shown in the simplified diagram of FIG. 2. The power amplifier 10 includes a switching device 12 and a load network 20. DC power is supplied to the device 12 via a choke 14. The network includes a simple filter 24 which is connected in series to an RL load, represented by LL 26, and RL 28, respectively. As a class E device, the filter acts as a short circuit at the fundamental frequency, and an open circuit at all harmonics. The inherent shunt capacitance, Cout, of the active device 12 (e.g. between the anode and cathode of a three terminal transistor) is incorporated into the network as all or part of capacitor Cs 22 which may include additional capacitance added by the designer. Thus, the impedance looking into the load network, Zin, is: at fo, Zin=(RL+jxcfx890LL)||(1/jxcfx890CS) which if properly designed is a substantially inductive load (i.e. a load consisting of both a resistance and an inductance), i.e. Zin=Reff+jxcfx890Leff; and at all harmonic overtones, Zin=1/jxcfx89CS. The inductance of the fundamental frequency load, when properly sized relative to the capacitance CS and the effective load resistance Reff, performs a phase correction of the fundamental frequency harmonic components, allowing the ZVS operation to be achieved.
Class F and Fxe2x88x921 Amplifiers
Class F is another well-known class of switching mode amplifiers. The class F amplifier derives its improved efficiency by using a multiple resonator load network to control the harmonic content of the active device""s output voltage and/or current waveforms. In realizing a class F circuit, the active device operates primarily as a switch and the load network, generally, is designed to yield short-circuit impedances at even harmonics of the fundamental frequency and to yield open-circuit impedances at odd harmonics of the fundamental frequency.
Efficient operation of a class F amplifier is realized when the output voltage of the active device (transistor) is driven rapidly from saturation (low resistance) to cutoff (high resistance) voltage. In operation, the combination of the active device and the output network produces a half sine wave current when the device is saturated. A high Q resonant circuit for all odd harmonics up to the Nth harmonic, often consisting of several parallel LC filters, makes possible odd harmonic components in the output voltage by providing high impedances to the active device at these frequencies. These odd harmonic voltages sum with the fundamental frequency output voltage to effectively flatten the output voltage waveform. This results in a combination of higher efficiency and higher power output. Additionally, resonant circuits are provided at all even harmonics up to the Nth harmonic to short circuit the active device at these frequencies, thereby allowing the current waveform to approximate a half-sinusoid, further increasing the efficiency without any decrease in output power. A high Q filter circuit is tuned to the fundamental frequency to reject harmonics at the load and yield a sinusoidal output signal. In this configuration, the device""s inherent parasitic capacitance must be kept small in order to avoid shorting the high impedance presented by the resonant circuit at the odd harmonics. Although this problem can be somewhat minimized by resonating this capacitance with the load network, this technique further increases the complexity of the network. Additionally, if the active device is to be driven very hard so that it switches very quickly, a large number of harmonics must be tuned in order to achieve the benefit of class F operation. As a consequence of these limitations, class F is normally used only in applications where the transistor speed is relatively slow compared to the frequency of operation and using relatively small (i.e. low capacitance) devices, so that only a few harmonics need be tuned and so that the effect of the capacitance is small.
A variation to the conventional class F amplifier is to invert the impedances at the harmonic overtones. Thus, the load network is designed to yield open circuit impedances at every even harmonic up to the Nth harmonic and short circuit impedances at every odd harmonic up to the Nth harmonic. Such an amplifier is called the inverse class-F, or class Fxe2x88x921 amplifier and one implementation is shown schematically in FIG. 3. In particular, this class Fxe2x88x921 amplifier 40 includes a switching device 42 and load network 50 that comprises a filter 46 in series with the output of the switch and the resistive load 52 and a second filter 48 in parallel with the load 52. The series filter 46 presents relatively open circuit impedances for even harmonics and short circuit impedances for all other harmonics. The parallel filter 48 presents relatively short circuit impedances for all odd harmonics and open circuit impedances otherwise. Thus, the impedance looking into the load network, Zin, is: at fo, Zin=RL; at all even harmonics Zin=∞(open); and at all odd harmonics Zin=0 (short). This amplifier class has many of the benefits of class F, and additionally has the property of near-ZVS operation, although this quality is difficult to achieve in the presence of a large parasitic device capacitance Cout. Although class Fxe2x88x921 has been largely ignored for many years, several recent works have shown that this class of operation compares favorably to class F using modern solid-state devices.
When class E and F power amplifier performances are compared, a significant advantage of a class E amplifier over a class F amplifier is its circuit topology, which incorporates the switching device output parasitic capacitance as part of its circuit. Thus, class E amplifiers do not lose power efficiency due to the charging and discharging of this parasitic capacitor as can occur in amplifier classes such as class F and class Fxe2x88x921 which do not account for the capacitor""s effect, nor do they require elaborate resonant circuits to reduce the effect of this capacitance. In addition, as seen above, the class E design is relatively simple, consisting of just a few components (at least one less filter than in the class F design). Unlike class F and Fxe2x88x921 designs, the class E design receives the full promised benefits of its operating class with this simple circuit, whereas the class F and Fxe2x88x921 approaches must incorporate increasingly larger numbers of circuit elements in order to approach the ideal class F performance. On the other hand, due to its anode (i.e., transistor drain or collector) voltage and current wave formats, class F amplifiers deliver significantly higher power and promise higher power-efficiency than class E amplifiers when they are using the same transistor under the same supply conditions. To gain this advantage, class F and Fxe2x88x921 circuits can be quite complex and can use many more components than class E devices.
Thus, it would be highly desirable to have a power amplifier capable of very efficiently providing high power at high frequencies and that incorporates some the best features of both class E and class F amplifiers.
The present invention, which addresses these needs, resides in a high efficiency switching power amplifier for amplifying a high frequency input signal having at least one fundamental frequency, and that is adapted to drive a load. The amplifier includes a high-speed active device and a hybrid class E/F load network. The active device comprises a switching component that operates substantially as a switch and a parasitic capacitance, Cout, in parallel with the switching component. The hybrid class E/F load network connected to the active device.
In one embodiment, the hybrid class E/F load network is configured to present to the switching component of the active device, at all harmonic frequencies substantially present in at least one of the voltage and current waveforms of the active device, a substantially inductive load at each fundamental frequency, a substantially open circuit at a predetermined number, NE, of even harmonic overtones for each fundamental frequency up to an Nth harmonic, a substantially short circuit at a predetermined number, NO, of odd harmonic overtones for each fundamental frequency up to an Nth harmonic, and a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic. In this embodiment, Nxe2x89xa73 and 1 less than NE+NOxe2x89xa6Nxe2x88x922. Thus, the amplifier has some characteristics of both a class E and class F amplifier. In a more specific example, if NE=1, then NO greater than 0.
More specifically, the load network includes a two port filter network having an input port and an output port, the input port being connected to the active device in parallel with the parasitic output capacitance Cout, and the output port being connected to the load. The load network may also be configured to provide wideband tuning of an input signal having a fundamental frequency range from f1 to f2, where f2 less than 3f1.
In another broad implementation of the present invention, the hybrid class E/F load network connected to the active device is configured to present to the active device a substantially inductive load at the fundamental frequency of operation, a substantially open circuit at a predetermined number of even harmonic overtones of the fundamental frequency, a substantially short circuit at a predetermined number of odd harmonic overtones of the fundamental frequency, and a substantially capacitive impedance load at the remaining harmonic overtones.
In yet another implementation of the present invention, the hybrid class E/F load network is configured to present to the switching component, at all harmonic frequencies that are substantially present in at least one of the voltage and current waveforms of the active device, a substantially inductive load at each fundamental frequency of operation that results in substantially zero-voltage-switching (ZVS) operation of the active device, impedances substantially larger in magnitude than 1(2xcfx80fCS) a at a predetermined number, NE, of even harmonic overtones of each fundamental frequency, impedances substantially smaller in magnitude than 1/(2xcfx80fCS) at a predetermined number, NO, of odd harmonic overtones of each fundamental frequency, and an impedance substantially equal to 1/jxcfx89CS at the remaining harmonic overtones of each fundamental frequency. CS=Cout+Cadded, where Caddedxe2x89xa70, and NExe2x89xa70, NOxe2x89xa70, and the total number of tuned harmonic overtones, NE+NO, is at least one and less than the total number of harmonic overtone frequencies substantially present in the active device""s at least one of voltage and current waveforms. Since the network need not operate to provide substantially open and short circuits, as in the prior examples, the network can be simplified to a significant degree.
In yet another implementation of the present invention, a multiple active device high efficiency switching power amplifier for amplifying a high frequency input signal having at least one fundamental frequency and adapted to drive a load, is disclosed. In this case, a first high-speed active device having a parasitic output capacitance, Cout1, and adapted to operate substantially as a switch, and a second high-speed active device having a parasitic output capacitance, Cout2 and adapted to operate substantially as a switch, are provided together with a hybrid three-port class E/F load network. The network has a first port connected to the first active device, a second port connected to the second active device, and a third port connected to the load, such that when the first and second active devices are driven in a push-pull configuration, the network presents to the switching component an effective input impedance that provides a substantially inductive load in series with the substantially resistive load at all fundamental frequencies; a substantially open circuit at one or more even harmonics for each fundamental frequency up to an Nth harmonic, a substantially short circuit at one or more odd harmonics for each fundamental frequency up to an Nth harmonic, and a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic.
In a more detailed implementation of this push-pull amplifier, the amplifier further includes a transformer connected to the outputs of the two active devices and the load such that the load is dc isolated from the outputs of the two active devices via the transformer.
In a detailed embodiment of one aspect of the present invention, a quasi-class E/F3 high efficiency amplifier for amplifying an input signal having at least one fundamental frequency and adapted to drive a load is disclosed. This amplifier includes a high speed active device that comprises a switching component that operates substantially as a switch and a parasitic capacitance, Cout, in parallel with the switching component and an LC parallel tank circuit that is resonant at the second harmonic of the fundamental frequency. The active device is connected in series to the load through the LC parallel tank circuit.
A method of amplifying an RF signal with an active device switch is also disclosed. The method includes amplifying the signal with an active device that comprises a switching component that operates substantially as a switch and a parasitic capacitance, Cout, in parallel with the switching component. The method includes tuning the amplified signal to provide a substantially inductive load to the switching component at the fundamental frequency, tuning the amplified signal to provide a substantially open circuit to the active device at selected even harmonic overtones, tuning the amplified signal to provide a substantially short circuit to the active device at selected odd harmonic overtones; and providing substantially capacitive loading to the active device for the non-selected harmonic overtones.
Several detailed implementations of the hybrid class E/F load network of the amplifier of the present invention are disclosed. In one embodiment, the network is configured to present to the switching component, at all harmonic frequencies substantially present in at least one of the voltage and current waveforms of the active device, a substantially inductive load at each fundamental frequency, a substantially open circuit at the 2nd harmonic, and a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic, where Nxe2x89xa73.
In an alternative implementation, the network is configured to present to the switching component, a substantially inductive load at each fundamental frequency; a substantially short circuit at the 3rd harmonic, and a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic where Nxe2x89xa73.
In a third detailed implementation, the hybrid class E/F load network is configured to present to the switching component a substantially inductive load at each fundamental frequency, a substantially short circuit at the 3rd harmonic, a substantially open circuit at the 2nd harmonic, and a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic, where Nxe2x89xa74.
In a fourth detailed embodiment, the hybrid class E/F load network is configured to present to the switching component a substantially inductive load at each fundamental frequency a substantially open circuit at the 4th harmonic, and a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic, where Nxe2x89xa74.
In a fifth detailed embodiment, the hybrid class E/F load network is configured to present to the switching component a substantially inductive load at each fundamental frequency a substantially open circuit at the 2nd and 4th harmonics, and a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic, where Nxe2x89xa74.
In a sixth embodiment, the hybrid class E/F load network is configured to present to the switching component a substantially inductive load at each fundamental frequency a substantially short circuit at the 3rd harmonic, a substantially open circuit at the 4th harmonic, and a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic, where Nxe2x89xa74.
In a seventh detailed embodiment, the hybrid class E/F load network is configured to present to the switching component a substantially inductive load at each fundamental frequency a substantially short circuit at the 3rd harmonic, a substantially open circuit at the 2nd and 4th harmonics, and a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic, where N xe2x89xa75.
In an eighth detailed embodiment, the hybrid class E/F load network is configured to present to the switching component a substantially inductive load at each fundamental frequency; a substantially short circuit at all odd harmonic overtones up to an Nth harmonic, a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic, where Nxe2x89xa75.
In a ninth disclosed detailed embodiment, the hybrid class E/F load network is configured to present to the switching component a substantially inductive load at each fundamental frequency a substantially short circuit at all odd harmonic overtones up to an Nth harmonic, a substantially open circuit at a predetermined number, NE, of even harmonic overtones for each fundamental frequency up to an Nth harmonic, a substantially capacitive impedance load at the remaining harmonic overtones, up to an Nth harmonic, where Nxe2x89xa75 and 0 less than NExe2x89xa6(Nxe2x88x922)/2.
Other features and advantages of the present invention will become apparent from the following detailed description, taken in conjunction with the accompanying drawings, which illustrate, by way of example, the principles of the invention.